On Tue, Apr 17, 2018 at 1:28 PM, Jens Axboe <axboe@xxxxxxxxx> wrote:
> It has to be the latter bfqq->dispatched increment, as those are
> transient (and bfqd is not).
Yeah, and I see a lot of comments around the lifetime of rq and bfqq,
so I assume something is not being locked correctly.
#define RQ_BFQQ(rq) ((rq)->elv.priv[1])
static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx)
{
struct bfq_data *bfqd = hctx->queue->elevator->elevator_data;
struct request *rq = NULL;
struct bfq_queue *bfqq = NULL;
if (!list_empty(&bfqd->dispatch)) {
rq = list_first_entry(&bfqd->dispatch, struct request,
queuelist);
list_del_init(&rq->queuelist);
bfqq = RQ_BFQQ(rq);
if (bfqq) {
/*
* Increment counters here, because this
* dispatch does not follow the standard
* dispatch flow (where counters are
* incremented)
*/
bfqq->dispatched++;
...
I see elv.priv[1] assignments made in a few places -- is it possible
there is some kind of uninitialized-but-not-NULL state that can leak
in there?
bfq_prepare_request() assigns elv.priv[1], and bfq_insert_request()
only checks that it's non-NULL (if at all) in one case. Can
bfq_insert_request() get called without bfq_prepare_request() being
called first?
-Kees
--
Kees Cook
Pixel Security
